Research funding awarded to faculty
Please join us in congratulating the following faculty members who were awarded funding for their research projects in 2023 and 2024! We also heartily thank the department and university staff who assist in seeking this funding and administering it when it arrives, in particular Christine Bishop.
Oksana Chkrebtii
Probabilistic Finite Elements for Uncertainty Quantification in Engineering and the
Natural Sciences. College of Arts and Sciences (ASC) Natural and Mathematical Sciences (NMS) Exploration Grant
Numerical methods are essential for studying the spatio-temporal evolution of physical systems defined implicitly by partial differential equations (PDEs) without a closed-form solution. However, small local errors in finite-dimensional solution approximations can propagate into large deviations from the true underlying model trajectory, impacting the validity of statistical inference and prediction for these dynamical systems. Incorporating a probability model of discretization uncertainty in the posterior hierarchy for unknown model parameters has recently provided a powerful new tool to alleviate inferential bias due to this systematic model misspecification. The overall objective of this proposal is to significantly extend this probabilistic numerical formalism to PDEs defined on irregular domains.
Sebastian Kurtek
Stochastic shape processes and inference, National Science Foundation Division of Mathematical Sciences. PI in collaboration with Anuj Srivastava, Karthik Bharath, and Ian Jermyn
The intimate link between form, or shape, and function is ubiquitous in science. In biology, for instance, the shapes of biological components are pivotal in understanding patterns of normal behavior and growth; a notable example is protein shape, which contributes to our understanding of protein function and classification. This project, led by a team of investigators from the USA and the UK, will develop ways of modeling how biological and other shapes change with time, using formal statistical frameworks that capture not only the changes themselves, but how these changes vary across objects and populations. This will enable the study of the link between form and function in all its variability. As example applications, the project will develop models for changes in cell morphology and topology during motility and division, and changes in human posture during various activities, facilitating the exploration of scientific questions such as how and why cell division fails, or how to improve human postures in factory tasks. These are proofs of concept, but the methods themselves will have much wider applicability. This project will thus not only progress the science of shape analysis and the specific applications studied; it will have broader downstream impacts on a range of scientific application domains, providing practitioners with general and useful tools.
While there are several approaches for representing and analyzing static shapes, encompassing curves, surfaces, and complex structures like trees and shape graphs, the statistical modeling and analysis of dynamic shapes has received limited attention. Mathematically, shapes are elements of quotient spaces of nonlinear manifolds, and shape changes can be modeled as stochastic processes, termed shape processes, on these complex spaces. The primary challenges lie in adapting classical modeling concepts to the nonlinear geometry of shape spaces and in developing efficient statistical tools for computation and inference in such very high-dimensional, nonlinear settings. The project consists of three thrust areas, dealing with combinations of discrete and continuous time, and discrete and continuous representations of shape, with a particular emphasis on the issues raised by topology changes. The key idea is to integrate spatiotemporal registration of objects and their evolution into the statistical formulation, rather than treating them as pre-processing steps. This project will specifically add to the current state-of-the-art in topic areas such as stochastic differential equations on shape manifolds, time series models for shapes, shape-based functional data analysis, and modeling and inference on infinite-dimensional shape spaces.
Shili Lin
Prenatal immunomonitoring in spontaneous preterm birth prevention (PROMIS), Eunice Kennedy Shriver National Institute of Child Health and Human Development. Co-PI
in collaboration with Amy Mackos, Ashley Felix, Maged Constantine, and Shannon Gillespie
Every year, more than 15M babies are born too soon, resulting in more than 1M deaths. In fact, preterm birth (PTB) affects approximately 1 in 10 births across the globe, including 1 in 10 US births. Despite this, methods for the prediction and prevention of PTB are sorely lacking. Of note, more than 60% of PTBs are spontaneous (sPTBs), brought about by premature contractions or membrane rupture. The most common finding in sPTB is significant inflammation among the expectant mother. Yet, levels of circulating and even localized inflammatory markers during pregnancy fail to clinically predict who will versus will not progress to future sPTB. Similarly, antibiotics for asymptomatic infections and anti-inflammatories do not reduce risk for sPTB and are therefore not recommended for broad administration. To address such deficits, we developed a novel immunomonitoring method for use in prenatal care with the goal of predicting future sPTB and informing allocation of targeted, preventive interventions. In other words, we aimed to diagnose and address the immunopathologies driving sPTB risk. We have now shown that our novel prenatal immunomonitoring method can predict future birth timing in low-risk cohorts and future sPTB in moderate to high-risk cohorts, outperforming all available methods. In addition, our ex vivo experiments showed considerable inter-individual variation in the immunomodulatory effects of progesterone on immune function during pregnancy, which may explain why the drug is beneficial for some but not others. Thus, quantifying these immunomodulatory effects may provide a direct avenue toward targeted prevention of sPTB using a drug that is known to be safe during pregnancy. In the “Prenatal Immunomonitoring in Spontaneous Preterm Birth Prevention (PROMIS)” study, we propose to refine, expand, and clinically validate our prenatal immunomonitoring methods in a large, diverse clinical cohort. Our long-term goal is to save lives by predicting and preventing future sPTB. Our central hypothesis is that this can be accomplished using our novel prenatal immunomonitoring methods, which aim to diagnose and address the drivers of sPTB risk. To test this hypothesis, we’ll enroll a diverse cohort presenting for the prenatal care of singleton pregnancy, with early pregnancy patient-oriented data and biospecimen collection, mid-pregnancy cervical length measurement, and post-birth medical record review. We’ll test a prenatal immunomonitoring algorithm for the prediction of future sPTB. We’ll characterize the immunomodulatory effects of progesterone and examine the association between this profile and future sPTB. We expect the PROMIS study to produce novel insights into the pathogenesis, prediction, and prevention of sPTB. Importantly, risk prediction and targeted prevention go hand-in-hand, making advancements in one area dependent upon our capacity to advance the other. The potential impact of this study lies in its potential to identify sPTB risk using clinically feasible methods AND characterize risk phenotypes in a manner that allows us to address them. Thus, this project could provide unprecedented opportunity to predict AND prevent sPTB.
Yoon Lee
Uncovering a conserved role of cell polarity signaling in cellular aging using budding yeast and hematopoietic stem cell models, The Ohio State University President’s Research Excellence (PRE) program. In collaboration with Hay-Oak Park and Brad Blaser
This proposal explores the interplay between conserved cell polarity signaling and aging by single-cell imaging and single-cell transcriptomics. Through the team’s interdisciplinary expertise, the project will address how Cdc42 and its effectors control lifespan in yeast and hematopoietic stem cells.
Steven MacEachern
Robust and efficient Bayesian inference for misspecified and underspecified models, National Science Foundation Division of Mathematical Sciences. PI in collaboration with Ju Hee Lee and Hang Joon Kim
This research project aims to improve data-driven modelling and decision-making. Its focus is on the development of Bayesian methods for low-information settings. Bayesian methods have proven to be tremendously successful in high-information settings where data is of high-quality, the scientific/business background that has generated the data is well-understood, and clear questions are asked. This project will develop a suite of Bayesian methods designed for low-information settings, including those where (i) the data show particular types of deficiencies, such as a preponderance of outlying or ?bad data?, (ii) a limited conceptual understanding of the phenomenon under study leads to a model that leaves a substantial gap between model and reality, producing a misspecified model or a model that is not fully specified, and (iii) when there is a shortage of data, so that the model captures only a very simplified version of reality. The new methods will expand the scope of Bayesian applications, with attention to problems in biomedical applications and psychology. The project will provide training for the next generation of data scientists.
This project has two main threads. For the first, the project will develop diagnostics that allow the analyst to assess the adequacy of portions of a posited model. Such assessments point the way toward elaborations that will bring the model closer to reality, improving the full collection of inferences. These assessments will also highlight limitations of the model, enabling the analyst to know when to make a decision and when to refrain from making one. The second thread will explore the use of sample-size adaptive loss functions for modelling and for inference. Adaptive loss functions have been used by classical statisticians to improve inference by exploiting the bias-variance tradeoff. This thread will blend adaptivity with Bayesian methods. This will robustify inference by providing smoother likelihoods for small and moderate sample sizes and by relying on smoother inference functions when the sample size is limited.
Omer Ozturk
Further adoption of ranked-set and judgement poststratification methodologies for treatment allocation and sampling designs in field experimentation. University of Adelaide.
The Biometry Hub is a consortium consisting of three universities in Australia—University of Adelaide, Curtin University, and University of Queensland. The hub secured a five-year research grant from the Grain Research Development Company (GRDC) in Australia. Omer Ozturk is a member of the University of Adelaide team.
Subhadeep Paul and Arnab Auddy
Transfer, Federated, and Private Statistical Learning for Trustworthy AI. College of Arts and Sciences (ASC) Natural and Mathematical Sciences (NMS) Exploration Grant.
See the research spotlight elsewhere in this newsletter for details.
David Sivakoff
Nucleation, excitation and annihilation in stochastic systems, The Simons Foundation
Each year, the Simons Foundation’s Mathematics and Physical Sciences division invites applications for its Travel Support for Mathematicians program, which is intended to stimulate collaboration in the field of mathematics primarily through the funding of travel and related expenditures.
Asuman Turkmen
Using machine learning and animal models to reveal bacterial subnetworks essential for development within complex gut microbiomes. National Science Foundation Directorate for Biological Sciences. Co-PI in collaboration with Zakee Sabree and Tanya Berger-Wolf.
Multicellular animals emerged into a microbial world and, many animals, including humans, maintain gut microbiomes that are complex microbial communities that they require for normal development and growth. The enormous species and functional diversity comprising these microbiomes confound efforts to link microbiome composition to specific healthy host phenotypes. The team will introduce random sub-samples of the total gut microbiome to a germ-free host and screen for those capable of resolving many of the growth and developmental deficiencies associated with being germ-free. Next, machine learning (ML) approaches will identify bacterial species that are consistently associated with promoting healthy host outcomes. Finally, the team will construct and introduce synthetic microbiomes comprised of bacterial species recommended by the ML models to germ-free animals to validate their predictions. Ultimately, this effort will identify specific bacterial lineages that are integral to animal growth, development and evolution. This interdisciplinary project leverages legacy and cutting-edge technologies to address what aspects of gut microbiome composition are essential for positive host outcomes. Additionally, this project will demonstrate the power of low-cost/high-replicate model systems and predictive modeling for rapid hypothesis generation and testing. Finally, investigators and postdoctoral scientists doing microbiome sciences will be afforded opportunities to recruit next-gen microbiome scientists from HBCUs with the goal of building a diverse microbiome science workforce. Further, these participants will obtain training in mentoring across different positionalities to enable them to build productive, long-term professional relationships with their mentees.
Host-associated bacteria are inextricably involved in the life history and evolution of metazoans. This research project builds on emerging evidence that suggests composition of gut microbiota (i.e. species/functional diversity) have large effects on host animal growth and development, and these effects are realized at several levels of biological organization (i.e. gene network expression, cellular proliferation and tissue differentiation, organismal body size and maturation). Many animals, including mammals, harbor species-rich and functionally complex gut microbiomes and identifying bacterial lineages within those complex communities that are critical for animal growth and development presents challenges that can be addressed through interdisciplinary approaches. Specifically, machine learning approaches will be used to integrate high-replicate multi-omics data from the gut microbiome and its host as well as host developmental, physiological and gastric histological data from several well-defined microbiome perturbations to infer bacterial species sub-networks that are consistently associated with normal host growth and development. The research team will test the hypothesis that these networks are host-supportive by constructing synthetic microbiomes comprised of predicted species in axenic juvenile hosts and track their development. This interdisciplinary approach leverages molecular and microbiological approaches, legacy (i.e. random forest) and cutting edge (i.e. convolution neural networks with triplet loss) machine learning tools, and an invertebrate animal model that normally harbors a complex gut microbiome and can easily be reared axenically without the use of antibiotics (i.e. Periplaneta americana) to shed light on the role of gut microbiota on host growth and development.
Xinyi Xu
Poly-matching causal inference for assessing multiple acute medical managements of pediatric traumatic brain injuries, Eunice Kennedy Shriver National Institute of Child Health and Human Development. In collaboration with Henry Xiang, Jonathan Groner and Bo Lu
Clinical effectiveness research (CER) plays a central role in research related to emergency medical services for children (EMSC). It is the conscientious use of the best available evidence in evaluating interventions, broadly defined as medical treatments, health policies, or practice patterns, which could lead to improvements in health care quality and patient outcomes. Observational data are more often used in the evaluation of healthcare systems or complex clinical practice than randomized controlled trials (RCTs), due to practical or ethical reasons. However, causal inference with observational data faces challenges: (1) Important covariates may be distributed differently between treatment options; (2) Conventional statistical analysis lacks control for unmeasured confounding. When the intervention is dichotomous, propensity score based adjustment is widely used to reduce the confounding bias introduced by observed covariates, through matching, stratification or weighting. Matching is a popular choice among researchers, as it creates data structures similar to RCTs, is easy to interpret, and robust to misspecifications in outcome modeling. But there is a critical methodological gap hindering the use of matching design when there are multiple (more than two) treatment options. This is due to the lack of good matching algorithms to generate well matched sets and the increased complexity of post-matching inference. Our overarching goal is to develop a statistically valid matching design (referred to as PMD) and subsequent causal inference procedures for use with complex observational healthcare databases, where there are multiple treatment arms or treatments over multiple time points. Specific aims: (1) Devise an innovative PMD for studies with multiple treatment arms or treatments over time; Develop causal inference strategies for PMD based on the potential outcome framework and sensitivity analysis strategies for assessing the unmeasured confounding effect; (2) Evaluate causal mortality impact of severe TBI (sTBI) patients who received trauma care at different type of trauma centers (PL1-level 1 pediatric, AL1-level 1 adult, and ML1-level 1 mixed trauma centers; PL2- level 2 pediatric, AL2-level 2 adult, and ML2-level 2 mixed trauma centers); (3) Assess the effectiveness of 4 Tier 1-2 medical management/therapies (ICP monitoring, head CT scan, cerebrospinal fluid drainage, decompressive craniectomy) on sTBI patient mortality; (4): Evaluate compliance with a Centers for Disease Control and Prevention (CDC) head CT guideline for mild TBI (mTBI) patients by different types of hospitals. This study is expected to fill a critical gap in EMSC research by extending the commonly used dichotomous matching design to complex observational studies with multiple treatment groups. This project is significant and our proposed methods are innovative as they include both observed confounding adjustment and unmeasured confounding assessment. We envision that this general-purpose methodology will be widely applicable and can benefit government agencies, policy makers, and social and health science researchers, where observational data are often utilized for comparative outcomes research and program/policy evaluation.
Yuan Zhang
U-statistic Reduction with Accurate Risk Control
U-statistics are a class of statistics that play central roles in many modern statistical learning tools. However, the heavy computational cost haunts their practical application. Despite considerable research efforts since the 1970s towards computational reduction for U-statistics, there exists little study on the important problem of accurate risk control in statistical inference for reduced U-statistics. Also, how computational speed trades off with risk control accuracy remains uncharacterized. This project will bridge this significant gap, providing the urgently needed infrastructual techniques that enable statisticians to securely scale up their U-statistic-based learning tools. The results of this project will provide profound benefits to a wide spectrum of research areas and applications, including nonparametric statistics, machine learning, sociology, computer vision and biomedical sciences. The project will also provide research training for graduate students.
This project will study both classical "noiseless" U-statistics and network U-statistics, an important subset of noisy U-statistics. The research will introduce innovative theoretical analysis techniques that lead to a sharp characterization of computational-statistical trade-offs and formulate new methods outperforming existing ones based on resampling and subsampling. The project aims to establish a general framework for principled analysis of different popular U-statistic reduction schemes. The research findings will be summarized into practical, step-by-step guides for easy implementation and tuning. The PI's team will also develop and disseminate user-friendly software for public users.
Grant abstracts taken from NSF Awards Search or NIH RePORTER, and university press releases